18.034 Honors Differential Equations - MIT · PDF file · 2017-12-28Separable...
Click here to load reader
Transcript of 18.034 Honors Differential Equations - MIT · PDF file · 2017-12-28Separable...
MIT OpenCourseWarehttp://ocw.mit.edu
18.034 Honors Differential Equations Spring 2009
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
� �
LECTURE 4. SEPARABLE EQUATIONS
Separable equations. Separable equations are differential equations of the form
dy f(x)(4.1) = .
dx g(y)
For example, x + yy� = 0 and y� = y2 − 1. A separable equation (4.1) can be written in the differential form as
(4.2) f(x)dx = g(y)dy.
Then, it can be solved formally by integrating both sides of (4.2). We state and prove the rigorous theory of local solutions for (4.2) (and hence (4.1)).
Theorem 4.1. Let f(x) and g(x) be continuous in the rectangle R = {(x, y) : a < x < b, c < y < d}. In addition, if f and g do not vanish simultaneously at any point of R, then (4.2) has one and only one solution through each point (x0, y0) ∈ R. The solution is given by
x y
(4.3) f(x)dx = g(y)dy x0 y0
It is essential that f and g do not vanish simultaneously. For example, xdx = −ydy has no solution through the origin.
Proof. Note that f(x) =� 0 for a < x < b or g(y) = 0 � for c < y < d. Without loss of generality, we assume g(y) > 0 for c < y < d.
Let � � x y
F (x) = f(x)dx, G(x) = g(y)dy x0 y0
so that (4.3) becomes
(4.4) F (x) = G(x).
Since G�(y) = g(y) > 0 in R, by the inverse function theorem, G−1 exists and (4.4) can be written as y = G−1(F (x)). That means, dy exists. Then, by differentiating (4.4), we get dx
F �(x) = G�(x) dy
, or f(x) = g(y) dy
. dx dx
This implies (4.2). Moreover, (4.3) gives the initial condition that y = y0 when x = x0. To prove the uniqueness, let y be one solution of (4.2) and z be another solution with the same
initial condition. Under the hypothesis, the equation
dz =
f(x) dx g(z)
implies that dz/dx exists for any (x, z) ∈ R. Let
u = G(y), v = G(z).
Then, du dy dy
= G�(y) = g(y) = f(x). dx dx dx
1
� �
dzSimilarly, = f(x). Since u and v have the same derivative, they differ by a constant. On the
dx other hand, the initial conditions for u and v at x0 agree. Therefore, u = v everywhere in R. This completes the proof. �
Example 4.2. Consider the initial value problem
(4.5) dy
= 1 + y 2 , y(0) = 1. dx
Separating the variables, we write the differential equation as dy
= dx.1 + y2
Since the constant function never vanishes, upon integration and evaluation, we obtain
tan−1 y = x + c, tan−1 1 = c.
Therefore, the (unique) solution of (4.5) is y = tan(x + π/4). The same result is obtained by integrating between corresponding limits
y dy x
= dx.1 + y2
1 0
Orthogonal trajectories. If two families of curves are such that every curve of one family intersects the curves of the other family at a right angle, then we say that the two families are orthogonal trajectories of each other. For example, the coordinate lines:
x = c1, y = c2
in a Cartesian coordinate system form a set of orthogonal trajectories. Another example is the circles and radial lines
r = c1, θ = c2
in a polar coordinate system. Suppose a curve in the (x, y)-plane is such that the tangent at a point (x, y) on it makes an angle
φ with the x-axis. The orthogonal trajectory through the same point (x, y) then makes an angle φ + π/2 with the x-axis. Since
1tan(φ + π/2) = − cot φ = −
tan φ
dy dx dxand since the slope of the curve is
dx = tan φ, we should replace
dy by −
dy in the differential
equation for the original family to get the differential equation for the orthogonal trajectories.
Example 4.3. We consider the family of circles
(4.6) x 2 + y 2 = cx
tangent to the y axis. By differentiating (4.6) and by eliminating c, we obtain a differential equation
dy 2 dyx 2 + y 2 = 2x + 2xy
dx, or y − x 2 = 2xy
dx that the family of curves (4.6) satisfies. Replace dy/dx by −dx/dy we get the equation of the orthogonal trajectories
dx y 2 − x 2 = −2xy
dy.
Lecture 4 2 18.034 Spring 2009
We write it in differential form as
2xydx − x 2dy + y 2dy = 0.
Multiplying by 1/y2 then gives∗ � �2x
d + dy = 0, y
2xand hence, + y = c. We arrange it into
y
x 2 + y 2 = cy,
which represents a family of circles tangent to the x-axis. Although the analytical steps require x = 0� , y = 0� , and y2 =� x2, the final result is valid without
these restrictions. The quantity c defined by x2 + y2 = cx is constant on the original curves of the family, but not
on the orthogonal trajectories. That is why c must be eliminated in the first step.
Exercise. Show that the orthogonal trajectories of the family of geometrically similar, coaxial ellipses
x 2 + my 2 = c, m > 0 mare given by y = ±|x| .
Exercise. Show that the solution curves of any separable equation y� = f(x)g(y) have as orthogonal trajectories the solution curves of the separable equation y� = −1/f(x)g(y).
∗This procedure makes the equation exact and the solution is defined implicitly. The factor 1/y2 is called a integrating factor. We will study exact differential equation more systematically later.
Lecture 4 3 18.034 Spring 2009